...
Dave Malham wrote:
Hi
On 28 June 2015 at 17:35, Dave Hunt <[email protected]>
wrote:
Hi,
I agree with Stefan.
One would have to undo a binaural encode to something else, then do a
dynamic head-tracked re-encode.
This can be "sort of" done by using blind source separation
techniques to
pull out the principle sources and their approximate positions then
panning
them into a B format form, adding the remainder as a more-or-less
diffuse
reverberant field. Note carefully I said "sort of" done - ok for
consumer
stuff (after all, mp3 is ok for consumer stuff) but not for anything
serious, except for use in compositions, where the choices are down
to the
composer.
Best
Another Dave
I wrote you would need the reconstruction of a real or "abstracted"
(say B format...) sound field from a binaural recording.
I don't believe there is a serious difference between "consumer stuff"
and "professional grade", in this case.
done by using blind source separation techniques to
pull out the principle sources and their approximate positions then
panning
them into a B format form
...
Hint: Would you not need some (individual) HRIR/HRTF measurements, to
apply "blind source separation"?
The biggest problem I see is that different directions could result in
similar (even identical) ear signals. In a mathematical sense, you
can't invert. To find (pragmatic) "predominant sound" directions, you
probably have too little and too ambiguous data to start with.
Yet another opinion about:
http://www.jeroenbreebaart.com/papers/jaes/jaes2008.pdf
see p. 11:
It is virtually impossible to undo or invert the encoderside
HRTF processing at the decoder (which is needed
in the second use case for loudspeaker playback).
Which basically confirms my position: "virtually impossible".
Anyway, who cares... ;-)
Best regards,
Stefan
I did once try to transform a binaural signal to first order B-Format
ambisonics with height using Max/MSP using a fairly simple
algorithm. It
sounded reasonably convincing, but was almost certainly not correct
in any
sense. It was just an attempt to use binaural files in a 1st order
ambisonic environment along with mono, stereo, MS etc.
Ciao,
Dave
_______________________________________________
Sursound mailing list
[email protected]
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit
account or options, view archives and so on.
